Upload folder using huggingface_hub

app.py

@@ -17,6 +17,7 @@ def run_quantization(
     layer_filter,
     exclude_layers_regex,
     full_precision_matrix_mult,
+    generic_text,
     hf_token
 ):
     if not all([source_repo, source_file, target_repo, target_filename_base]):
@@ -80,8 +81,11 @@ def run_quantization(
         if full_precision_matrix_mult:
             quant_args["full_precision_matrix_mult"] = True
 
-        if exclude_layers_regex:
-            quant_args["exclude_layers"] = exclude_layers_regex
+        if generic_text:
+            quant_args["generic_text"] = True
+
+        if exclude_layers_regex:
+            quant_args["exclude_layers"] = exclude_layers_regex
 
         yield f"Quantizing to {output_filename}...\nThis may take a few minutes.", gr.update(visible=False), gr.update(visible=False)
 
@@ -154,6 +158,7 @@ with gr.Blocks() as demo:
                 value="None", label="Model Layer Filter"
             )
             full_precision = gr.Checkbox(label="Full precision matrix multiplication", value=False)
+            generic_text = gr.Checkbox(label="Generic text model quantization", value=False)
 
             exclude_layers = gr.Textbox(label="Exclude Layers Regex (Optional)", placeholder="(substring_1|substring_2)")
 
@@ -183,6 +188,7 @@ with gr.Blocks() as demo:
         inputs=[
             source_repo, source_file, target_repo, target_file_base,
             quant_format, layer_filter, exclude_layers, full_precision,
+            generic_text,
             hf_token
         ],
         outputs=[status_text, output_link, output_file]